
import os
import warnings
import pprint
import pandas as pd
import numpy as np
import seaborn as sns
import matplotlib.pyplot as plt
import sklearn.metrics as metrics
from datetime import datetime
from matplotlib import gridspec
from sklearn.linear_model import LogisticRegression
from sklearn.model_selection import train_test_split
from sklearn.tree import DecisionTreeClassifier
from sklearn.preprocessing import OneHotEncoder
from sklearn.model_selection import GridSearchCV
from sklearn import tree
from os import system
from IPython.display import Image


warnings.filterwarnings('ignore')

main_data = pd.read_csv('My Bank.csv')

default_color_palette  = ["#03A9F4","#7C4DFF", "#FF5252", "#D50000", "#FF6F00", "#0288D1", "#7C4DFF"] 
stats_colors           = {'Mean':'#D50000', 'Mode':'#FF3D00', 'Median':'#2962FF'} # Set standard colors for mean, mode and median to use accross entire notebook.
gender_color_palette   = {"Female":"#E91E63", "Male":"#42A5F5"} # Gender based colors.


main_data.head(10)


main_data.drop("CUST_ID", axis=1, inplace=True)
main_data.info(10)

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 20000 entries, 0 to 19999
Data columns (total 25 columns):
 #   Column                    Non-Null Count  Dtype  
---  ------                    --------------  -----  
 0   TARGET                    20000 non-null  int64  
 1   AGE                       20000 non-null  int64  
 2   GENDER                    20000 non-null  object 
 3   BALANCE                   20000 non-null  float64
 4   OCCUPATION                20000 non-null  object 
 5   SCR                       20000 non-null  int64  
 6   HOLDING_PERIOD            20000 non-null  int64  
 7   ACC_TYPE                  20000 non-null  object 
 8   ACC_OP_DATE               20000 non-null  object 
 9   LEN_OF_RLTN_IN_MNTH       20000 non-null  int64  
 10  NO_OF_L_CR_TXNS           20000 non-null  int64  
 11  NO_OF_BR_CSH_WDL_DR_TXNS  20000 non-null  int64  
 12  NO_OF_ATM_DR_TXNS         20000 non-null  int64  
 13  NO_OF_NET_DR_TXNS         20000 non-null  int64  
 14  NO_OF_MOB_DR_TXNS         20000 non-null  int64  
 15  NO_OF_CHQ_DR_TXNS         20000 non-null  int64  
 16  FLG_HAS_CC                20000 non-null  int64  
 17  AMT_ATM_DR                20000 non-null  int64  
 18  AMT_BR_CSH_WDL_DR         20000 non-null  int64  
 19  AMT_CHQ_DR                20000 non-null  int64  
 20  AMT_NET_DR                20000 non-null  float64
 21  AMT_MOB_DR                20000 non-null  int64  
 22  FLG_HAS_ANY_CHGS          20000 non-null  int64  
 23  FLG_HAS_NOMINEE           20000 non-null  int64  
 24  FLG_HAS_OLD_LOAN          20000 non-null  int64  
dtypes: float64(2), int64(19), object(4)
memory usage: 3.8+ MB


'''
Input:
Pandas DataFrame

Output:
Displays DataFrame structure 
(columns, nulls and non nulls counts and percentage highlighing columns with most number of nulls)

Retunrs:
N/A
'''
def info(dataFrame):
    print(f"{dataFrame.shape[0]} Rows x {dataFrame.shape[1]} Columns")
    nulls_series             = dataFrame.isna().sum()         # Get a series counting number of empty values for each column
    nonnulls_series          = dataFrame.notnull().sum()      # Get a series counting number of non empty valuesfor each column 
    nulls_percentage         = ((nulls_series * 100)/(nulls_series + nonnulls_series)).astype(float)
    column_datatypes         = dataFrame.dtypes               # Get a series containing data types for each column 

    nulls_count_df = pd.concat(
        [nulls_series, nonnulls_series, nulls_percentage, column_datatypes],
        axis=1,
        keys=["Nulls", "Non-Nulls","Nulls %", "Type"], 
        sort=True)
    cm = sns.light_palette("red", as_cmap=True)
    display(nulls_count_df.style.background_gradient(cmap=cm, subset=pd.IndexSlice[:, ['Nulls %']]).format(formatter={('Nulls %'): "{:.2f}%"}))


info(main_data)

20000 Rows x 25 Columns


main_data[main_data.select_dtypes(include=['object']).columns.tolist()].head(10)


# pull out all features with less than or equal to five unique values
to_category_features = main_data.nunique()[main_data.nunique().values <= 5].index.values
main_data.nunique()[main_data.nunique().values <= 5].head(10)

TARGET              2
GENDER              3
OCCUPATION          4
ACC_TYPE            2
FLG_HAS_CC          2
FLG_HAS_ANY_CHGS    2
FLG_HAS_NOMINEE     2
FLG_HAS_OLD_LOAN    2
dtype: int64


# Explore types for the features pulled in the previous step
main_data[main_data.nunique()[main_data.nunique().values <= 5].index.to_list()].info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 20000 entries, 0 to 19999
Data columns (total 8 columns):
 #   Column            Non-Null Count  Dtype 
---  ------            --------------  ----- 
 0   TARGET            20000 non-null  int64 
 1   GENDER            20000 non-null  object
 2   OCCUPATION        20000 non-null  object
 3   ACC_TYPE          20000 non-null  object
 4   FLG_HAS_CC        20000 non-null  int64 
 5   FLG_HAS_ANY_CHGS  20000 non-null  int64 
 6   FLG_HAS_NOMINEE   20000 non-null  int64 
 7   FLG_HAS_OLD_LOAN  20000 non-null  int64 
dtypes: int64(5), object(3)
memory usage: 1.2+ MB


'''
Input:
Target type and a list of feature names.

Output:
Convert all features provided in 'column_names' to Target type provided in 'toType'

Returns:
modifies main original data frame and returns nothing.
'''
def ConvertColTo(toType, column_names=np.nan, df = np.nan):
    for col_name in column_names:
        if col_name != 'TARGET':
            if df != df:
                main_data[col_name] = main_data[col_name].astype(toType)
            else:
                df[col_name] = df[col_name].astype(toType)

ConvertColTo("category", to_category_features)
main_data.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 20000 entries, 0 to 19999
Data columns (total 25 columns):
 #   Column                    Non-Null Count  Dtype   
---  ------                    --------------  -----   
 0   TARGET                    20000 non-null  int64   
 1   AGE                       20000 non-null  int64   
 2   GENDER                    20000 non-null  category
 3   BALANCE                   20000 non-null  float64 
 4   OCCUPATION                20000 non-null  category
 5   SCR                       20000 non-null  int64   
 6   HOLDING_PERIOD            20000 non-null  int64   
 7   ACC_TYPE                  20000 non-null  category
 8   ACC_OP_DATE               20000 non-null  object  
 9   LEN_OF_RLTN_IN_MNTH       20000 non-null  int64   
 10  NO_OF_L_CR_TXNS           20000 non-null  int64   
 11  NO_OF_BR_CSH_WDL_DR_TXNS  20000 non-null  int64   
 12  NO_OF_ATM_DR_TXNS         20000 non-null  int64   
 13  NO_OF_NET_DR_TXNS         20000 non-null  int64   
 14  NO_OF_MOB_DR_TXNS         20000 non-null  int64   
 15  NO_OF_CHQ_DR_TXNS         20000 non-null  int64   
 16  FLG_HAS_CC                20000 non-null  category
 17  AMT_ATM_DR                20000 non-null  int64   
 18  AMT_BR_CSH_WDL_DR         20000 non-null  int64   
 19  AMT_CHQ_DR                20000 non-null  int64   
 20  AMT_NET_DR                20000 non-null  float64 
 21  AMT_MOB_DR                20000 non-null  int64   
 22  FLG_HAS_ANY_CHGS          20000 non-null  category
 23  FLG_HAS_NOMINEE           20000 non-null  category
 24  FLG_HAS_OLD_LOAN          20000 non-null  category
dtypes: category(7), float64(2), int64(15), object(1)
memory usage: 2.9+ MB


main_data['ACC_OP_DATE'] = pd.to_datetime(main_data['ACC_OP_DATE'])
main_data.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 20000 entries, 0 to 19999
Data columns (total 25 columns):
 #   Column                    Non-Null Count  Dtype         
---  ------                    --------------  -----         
 0   TARGET                    20000 non-null  int64         
 1   AGE                       20000 non-null  int64         
 2   GENDER                    20000 non-null  category      
 3   BALANCE                   20000 non-null  float64       
 4   OCCUPATION                20000 non-null  category      
 5   SCR                       20000 non-null  int64         
 6   HOLDING_PERIOD            20000 non-null  int64         
 7   ACC_TYPE                  20000 non-null  category      
 8   ACC_OP_DATE               20000 non-null  datetime64[ns]
 9   LEN_OF_RLTN_IN_MNTH       20000 non-null  int64         
 10  NO_OF_L_CR_TXNS           20000 non-null  int64         
 11  NO_OF_BR_CSH_WDL_DR_TXNS  20000 non-null  int64         
 12  NO_OF_ATM_DR_TXNS         20000 non-null  int64         
 13  NO_OF_NET_DR_TXNS         20000 non-null  int64         
 14  NO_OF_MOB_DR_TXNS         20000 non-null  int64         
 15  NO_OF_CHQ_DR_TXNS         20000 non-null  int64         
 16  FLG_HAS_CC                20000 non-null  category      
 17  AMT_ATM_DR                20000 non-null  int64         
 18  AMT_BR_CSH_WDL_DR         20000 non-null  int64         
 19  AMT_CHQ_DR                20000 non-null  int64         
 20  AMT_NET_DR                20000 non-null  float64       
 21  AMT_MOB_DR                20000 non-null  int64         
 22  FLG_HAS_ANY_CHGS          20000 non-null  category      
 23  FLG_HAS_NOMINEE           20000 non-null  category      
 24  FLG_HAS_OLD_LOAN          20000 non-null  category      
dtypes: category(7), datetime64[ns](1), float64(2), int64(15)
memory usage: 2.9 MB


main_data.describe().T


'''
Input:
N/A

Output:
Go through each categorial column and print unique values for that column.

Retunrs:
N/A
'''
def CountUniqueValues(col_names):
    for col_name in col_names:
        print(f"======================='{col_name}'==================")
        for unique_col_value in  main_data[col_name].unique().tolist():
            total_count = main_data[col_name].count()
            unique_values_count = main_data[main_data[col_name] == unique_col_value][col_name].count()
            percentage = str(round((unique_values_count/total_count) * 100, 2))
            print(f"{unique_col_value} \t: {unique_values_count} ({percentage}%)")
        print(f"=========================================================\n")

catgry_col_names = main_data.select_dtypes(include=['category']).columns.tolist()
CountUniqueValues(catgry_col_names)
CountUniqueValues(['TARGET'])

======================='GENDER'==================
M 	: 14376 (71.88%)
F 	: 5433 (27.16%)
O 	: 191 (0.96%)
=========================================================

======================='OCCUPATION'==================
SELF-EMP 	: 3568 (17.84%)
SAL 	: 5855 (29.28%)
PROF 	: 5417 (27.08%)
SENP 	: 5160 (25.8%)
=========================================================

======================='ACC_TYPE'==================
SA 	: 15759 (78.8%)
CA 	: 4241 (21.2%)
=========================================================

======================='FLG_HAS_CC'==================
0 	: 13892 (69.46%)
1 	: 6108 (30.54%)
=========================================================

======================='FLG_HAS_ANY_CHGS'==================
0 	: 17788 (88.94%)
1 	: 2212 (11.06%)
=========================================================

======================='FLG_HAS_NOMINEE'==================
1 	: 18023 (90.12%)
0 	: 1977 (9.88%)
=========================================================

======================='FLG_HAS_OLD_LOAN'==================
1 	: 9859 (49.3%)
0 	: 10141 (50.7%)
=========================================================

======================='TARGET'==================
0 	: 17488 (87.44%)
1 	: 2512 (12.56%)
=========================================================


main_data.nunique()

TARGET                         2
AGE                           35
GENDER                         3
BALANCE                     9924
OCCUPATION                     4
SCR                          896
HOLDING_PERIOD                31
ACC_TYPE                       2
ACC_OP_DATE                 4869
LEN_OF_RLTN_IN_MNTH          193
NO_OF_L_CR_TXNS               76
NO_OF_BR_CSH_WDL_DR_TXNS      16
NO_OF_ATM_DR_TXNS             26
NO_OF_NET_DR_TXNS             17
NO_OF_MOB_DR_TXNS              6
NO_OF_CHQ_DR_TXNS             16
FLG_HAS_CC                     2
AMT_ATM_DR                   646
AMT_BR_CSH_WDL_DR           7369
AMT_CHQ_DR                  4891
AMT_NET_DR                  4847
AMT_MOB_DR                  2312
FLG_HAS_ANY_CHGS               2
FLG_HAS_NOMINEE                2
FLG_HAS_OLD_LOAN               2
dtype: int64


# count the number of duplicates with 'TARGET' equal to Zero '0'.
main_data[main_data.duplicated(keep=False)][main_data['TARGET'] == 0].duplicated(keep=False).sum()

13044


main_data[main_data.duplicated(keep=False)][main_data['TARGET'] == 0].head(10)


# Drop duplicate rows with 'TARGET' equal to Zero '0'.
main_data = pd.concat([
        main_data[main_data.duplicated(keep=False)][main_data['TARGET'] == 0].drop_duplicates(),
        main_data[main_data['TARGET'] == 1]
    ])


CountUniqueValues(catgry_col_names)
CountUniqueValues(['TARGET'])

======================='GENDER'==================
M 	: 5302 (73.4%)
F 	: 1840 (25.47%)
O 	: 81 (1.12%)
=========================================================

======================='OCCUPATION'==================
SELF-EMP 	: 1488 (20.6%)
SAL 	: 2035 (28.17%)
PROF 	: 1919 (26.57%)
SENP 	: 1781 (24.66%)
=========================================================

======================='ACC_TYPE'==================
SA 	: 5567 (77.07%)
CA 	: 1656 (22.93%)
=========================================================

======================='FLG_HAS_CC'==================
0 	: 4779 (66.16%)
1 	: 2444 (33.84%)
=========================================================

======================='FLG_HAS_ANY_CHGS'==================
0 	: 6361 (88.07%)
1 	: 862 (11.93%)
=========================================================

======================='FLG_HAS_NOMINEE'==================
1 	: 6520 (90.27%)
0 	: 703 (9.73%)
=========================================================

======================='FLG_HAS_OLD_LOAN'==================
1 	: 3536 (48.95%)
0 	: 3687 (51.05%)
=========================================================

======================='TARGET'==================
0 	: 4711 (65.22%)
1 	: 2512 (34.78%)
=========================================================


'''
Input:
Axis, column name/x-axis, hue name

Output:
Displays a count plot.

Retunrs:
N/A
''' 
def HistBoxplot(box_chart_ax, hist_chart_ax, x_axis):
    sns.boxplot(
        data=main_data,
        x=main_data[x_axis],
        showmeans=True,
        ax=box_chart_ax)
    
    sns.histplot(
        data=main_data,
        x=main_data[x_axis],
        kde=True,
        ax=hist_chart_ax)
    
    hist_chart_ax.axvline(main_data[x_axis].mean(),                 # Get the mean of the values in the given column and draw a vertical line that cuts the chart on the mean value 
               color=stats_colors['Mean'],                                  # Use on of the colors predefined on this notebook
               label='Mean',                                                # Set the label to be diplayed on the legend
               linestyle="dashed");                                         # Make the line have dashes
    
    hist_chart_ax.axvline(main_data[x_axis].median(),               # Plot the median line on the chart.
               color=stats_colors['Median'],                                # Use on of the colors predefined on this notebook
               label='Median',                                              # Set the label to be diplayed on the legend
               linestyle="dashed");                                         # Make the line have dashes
    
    hist_chart_ax.axvline(main_data[x_axis].mode()[0],              # Plot the mode line on the chart.
               color=stats_colors['Mode'],                                  # Use on of the colors predefined on this notebook 
               label='Mode',                                                # Set the label to be diplayed on the legend
               linestyle="dashed");                                         # Make the line have dashes
    
    hist_chart_ax.legend(loc='upper right')

'''
Input:
N/A

Output:
Displays a a grid of [Boxplot x Distribution chart] for discrete features.

Returns:
N/A
''' 
def PlotHistBoxGrid():
    col_names = main_data.select_dtypes(include=['float', 'int64']).columns.tolist()
    print(col_names)

    col_index = 0
    for r in range(0, int(len(col_names)), 3):
        fig, (box, hist) = plt.subplots(
            nrows=2,                                                                # Number of rows of the grid
            ncols=3,                                                                # Number of columns of the grid.
            figsize=(15,4),
            gridspec_kw={"height_ratios" : (0.25,0.5)},
            constrained_layout=True)

        for index in range(0, 3):
            if col_index < int(len(col_names)):
                HistBoxplot(box.flat[index], hist.flat[index], col_names[col_index])
            col_index += 1
            
'''
Input:
Column name

Output:
A series containing interquatile range values

Retunrs:
A series containing interquatile range values
''' 
def Get_IQR(col_name):
    quartiles = np.quantile(main_data[col_name][main_data[col_name].notnull()], [.2, .75])
    power_4iqr = 4 * (quartiles[1] - quartiles[0])
    print(f'Q1 = {quartiles[0]}, Q3 = {quartiles[1]}, 4*IQR = {power_4iqr}')
    outlier_powers = main_data.loc[np.abs(main_data[col_name] - main_data[col_name].median()) > power_4iqr, col_name]
    print(outlier_powers.array) 
    
    return quartiles


PlotHistBoxGrid()

['TARGET', 'AGE', 'BALANCE', 'SCR', 'HOLDING_PERIOD', 'LEN_OF_RLTN_IN_MNTH', 'NO_OF_L_CR_TXNS', 'NO_OF_BR_CSH_WDL_DR_TXNS', 'NO_OF_ATM_DR_TXNS', 'NO_OF_NET_DR_TXNS', 'NO_OF_MOB_DR_TXNS', 'NO_OF_CHQ_DR_TXNS', 'AMT_ATM_DR', 'AMT_BR_CSH_WDL_DR', 'AMT_CHQ_DR', 'AMT_NET_DR', 'AMT_MOB_DR']


fig, ax = plt.subplots(figsize=(18,18)) 
sns.heatmap(data=main_data.corr(), annot=True, linewidths=.5, ax=ax)
plt.show()


main_data['BALANCE_log'] = (main_data['BALANCE'] + 1).transform(np.log)
main_data['NO_OF_CHQ_DR_TXNS_log'] = (main_data['NO_OF_CHQ_DR_TXNS'] + 1).transform(np.log)

main_data.drop([
    'BALANCE',
    'NO_OF_CHQ_DR_TXNS'
],
    axis=1,
    inplace=True)
main_data.info()

<class 'pandas.core.frame.DataFrame'>
Int64Index: 7223 entries, 0 to 19999
Data columns (total 25 columns):
 #   Column                    Non-Null Count  Dtype         
---  ------                    --------------  -----         
 0   TARGET                    7223 non-null   int64         
 1   AGE                       7223 non-null   int64         
 2   GENDER                    7223 non-null   category      
 3   OCCUPATION                7223 non-null   category      
 4   SCR                       7223 non-null   int64         
 5   HOLDING_PERIOD            7223 non-null   int64         
 6   ACC_TYPE                  7223 non-null   category      
 7   ACC_OP_DATE               7223 non-null   datetime64[ns]
 8   LEN_OF_RLTN_IN_MNTH       7223 non-null   int64         
 9   NO_OF_L_CR_TXNS           7223 non-null   int64         
 10  NO_OF_BR_CSH_WDL_DR_TXNS  7223 non-null   int64         
 11  NO_OF_ATM_DR_TXNS         7223 non-null   int64         
 12  NO_OF_NET_DR_TXNS         7223 non-null   int64         
 13  NO_OF_MOB_DR_TXNS         7223 non-null   int64         
 14  FLG_HAS_CC                7223 non-null   category      
 15  AMT_ATM_DR                7223 non-null   int64         
 16  AMT_BR_CSH_WDL_DR         7223 non-null   int64         
 17  AMT_CHQ_DR                7223 non-null   int64         
 18  AMT_NET_DR                7223 non-null   float64       
 19  AMT_MOB_DR                7223 non-null   int64         
 20  FLG_HAS_ANY_CHGS          7223 non-null   category      
 21  FLG_HAS_NOMINEE           7223 non-null   category      
 22  FLG_HAS_OLD_LOAN          7223 non-null   category      
 23  BALANCE_log               7223 non-null   float64       
 24  NO_OF_CHQ_DR_TXNS_log     7223 non-null   float64       
dtypes: category(7), datetime64[ns](1), float64(3), int64(14)
memory usage: 1.1 MB


fig, ax = plt.subplots(figsize=(18,18)) 
sns.heatmap(data=main_data.corr(), annot=True, linewidths=.5, ax=ax)
plt.show()


PlotHistBoxGrid()

['TARGET', 'AGE', 'SCR', 'HOLDING_PERIOD', 'LEN_OF_RLTN_IN_MNTH', 'NO_OF_L_CR_TXNS', 'NO_OF_BR_CSH_WDL_DR_TXNS', 'NO_OF_ATM_DR_TXNS', 'NO_OF_NET_DR_TXNS', 'NO_OF_MOB_DR_TXNS', 'AMT_ATM_DR', 'AMT_BR_CSH_WDL_DR', 'AMT_CHQ_DR', 'AMT_NET_DR', 'AMT_MOB_DR', 'BALANCE_log', 'NO_OF_CHQ_DR_TXNS_log']


previous_corr = [main_data.corr()[['TARGET']]]
col_names = main_data.select_dtypes(include=['float', 'int64']).columns.tolist()

main_data_copy = main_data.copy()
for col_name in col_names:
    print(f"\n\n==========================={col_name}====================\n")
    main_data[col_name] = np.where(
        main_data_copy[col_name] < Get_IQR(col_name)[0],
        main_data_copy[col_name].mode(),
        main_data_copy[col_name])
    
previous_corr.append(main_data_copy.corr()[['TARGET']])


===========================TARGET====================

Q1 = 0.0, Q3 = 1.0, 4*IQR = 4.0
<PandasArray>
[]
Length: 0, dtype: int64


===========================AGE====================

Q1 = 29.0, Q3 = 46.0, 4*IQR = 68.0
<PandasArray>
[]
Length: 0, dtype: int64


===========================SCR====================

Q1 = 212.0, Q3 = 659.0, 4*IQR = 1788.0
<PandasArray>
[]
Length: 0, dtype: int64


===========================HOLDING_PERIOD====================

Q1 = 6.0, Q3 = 21.0, 4*IQR = 60.0
<PandasArray>
[]
Length: 0, dtype: int64


===========================LEN_OF_RLTN_IN_MNTH====================

Q1 = 69.0, Q3 = 172.0, 4*IQR = 412.0
<PandasArray>
[]
Length: 0, dtype: int64


===========================NO_OF_L_CR_TXNS====================

Q1 = 5.0, Q3 = 16.0, 4*IQR = 44.0
<PandasArray>
[58, 72, 55, 60, 63, 57, 68, 65, 67, 73,
 ...
 66, 59, 67, 55, 64, 75, 60, 61, 58, 55]
Length: 225, dtype: int64


===========================NO_OF_BR_CSH_WDL_DR_TXNS====================

Q1 = 0.0, Q3 = 3.0, 4*IQR = 12.0
<PandasArray>
[14, 15, 14, 15, 14, 14, 15, 14, 14, 15, 15, 14, 14, 14, 14, 14, 14]
Length: 17, dtype: int64


===========================NO_OF_ATM_DR_TXNS====================

Q1 = 0.0, Q3 = 1.0, 4*IQR = 4.0
<PandasArray>
[10, 20, 11, 12, 13, 20, 14,  7, 12, 20, 10, 20, 17, 22,  6, 12, 16,  7,  8,
 20, 14, 18, 25, 11, 10, 16, 23, 21, 16,  8, 21, 15, 19, 15, 10,  6, 21, 10,
 20, 20,  8, 10, 10,  8, 23, 23, 15,  9, 20, 21, 20, 23,  8, 23, 23,  8,  6,
  8, 16]
Length: 59, dtype: int64


===========================NO_OF_NET_DR_TXNS====================

Q1 = 0.0, Q3 = 1.0, 4*IQR = 4.0
<PandasArray>
[ 7,  7,  6,  6,  6, 10,  6,  6, 19,  7,
 ...
  6, 18,  8,  6,  7,  6,  6,  8,  8,  6]
Length: 581, dtype: int64


===========================NO_OF_MOB_DR_TXNS====================

Q1 = 0.0, Q3 = 0.0, 4*IQR = 0.0
<PandasArray>
[1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
 ...
 1, 1, 1, 1, 1, 2, 1, 1, 2, 1]
Length: 1803, dtype: int64


===========================AMT_ATM_DR====================

Q1 = 0.0, Q3 = 16500.0, 4*IQR = 66000.0
<PandasArray>
[ 75900,  79900, 128400, 196600, 185900, 199100,  92700,  96100, 130100,
  95500,  90900,  74800, 107000, 147700,  74700,  74500, 117700, 199300,
  91900, 140900, 101300,  82200,  88800,  74900, 147000, 117200, 147000,
 101300,  91200,  82200,  88800, 152000,  95000,  82200, 152000,  78700,
 152000, 127200, 147000,  88800, 180200, 127200, 127200,  95000,  91200,
  90600,  78700]
Length: 47, dtype: int64


===========================AMT_BR_CSH_WDL_DR====================

Q1 = 0.0, Q3 = 688160.0, 4*IQR = 2752640.0
<PandasArray>
[]
Length: 0, dtype: int64


===========================AMT_CHQ_DR====================

Q1 = 0.0, Q3 = 76910.0, 4*IQR = 307640.0
<PandasArray>
[ 844830,  341900,  865910,  716480,  625310,  562570,  346850, 3511970,
  944110,  520790,
 ...
  465110,  597200,  471500,  984630,  979080,  754600,  670840, 1674070,
  698100,  617030]
Length: 653, dtype: int64


===========================AMT_NET_DR====================

Q1 = 0.0, Q3 = 504583.0, 4*IQR = 2018332.0
<PandasArray>
[]
Length: 0, dtype: float64


===========================AMT_MOB_DR====================

Q1 = 0.0, Q3 = 0.0, 4*IQR = 0.0
<PandasArray>
[ 71388, 170332, 171334,  52825,   2141, 198996, 177483,  65064, 171636,
  83531,
 ...
  96085, 127426, 104727,  20766, 118231, 119174, 104170, 147672,  18429,
 154075]
Length: 1803, dtype: int64


===========================BALANCE_log====================

Q1 = 10.624641599623047, Q3 = 13.295236663037764, 4*IQR = 10.68238025365887
<PandasArray>
[               0.0,                0.0, 0.5364933705145685,
                0.0,                0.0,                0.0,
                0.0,                0.0,                0.0,
  1.501852701754163,                0.0,                0.0,
                0.0,                0.0,                0.0,
                0.0,                0.0,                0.0,
                0.0,                0.0,                0.0,
                0.0,                0.0,                0.0,
 1.2237754316221157,                0.0, 0.6259384308664954,
 0.6259384308664954, 1.2237754316221157, 0.6259384308664954,
                0.0, 0.6259384308664954, 0.6259384308664954,
 0.6259384308664954, 1.1878434223960523]
Length: 35, dtype: float64


===========================NO_OF_CHQ_DR_TXNS_log====================

Q1 = 0.0, Q3 = 1.6094379124341003, 4*IQR = 6.437751649736401
<PandasArray>
[]
Length: 0, dtype: float64


fig, ax = plt.subplots(figsize=(10,10)) 
sns.heatmap(data=pd.concat(previous_corr, axis=1), annot=True, linewidths=.5, ax=ax)
plt.show()


oneHotCols = main_data.select_dtypes(include=['category']).columns.tolist()
main_data = pd.get_dummies(main_data, columns=oneHotCols, drop_first=True)

main_data.head(10)


type(main_data['ACC_OP_DATE'].to_list()[0])

main_data['ACC_OP_DATE_MONTH']     = main_data['ACC_OP_DATE'].dt.month
main_data['ACC_OP_DATE_DAYNAME']   = main_data['ACC_OP_DATE'].dt.day_name()
main_data['ACC_OP_DATE_AGE_IN_MONTHS']       = (datetime.today().year - main_data['ACC_OP_DATE'].dt.year) * 12 + datetime.today().month - main_data['ACC_OP_DATE'].dt.month 

main_data.drop(['ACC_OP_DATE'], axis=1, inplace=True)
main_data.info()

<class 'pandas.core.frame.DataFrame'>
Int64Index: 7223 entries, 0 to 19999
Data columns (total 30 columns):
 #   Column                     Non-Null Count  Dtype  
---  ------                     --------------  -----  
 0   TARGET                     7223 non-null   int64  
 1   AGE                        7223 non-null   int64  
 2   SCR                        7223 non-null   int64  
 3   HOLDING_PERIOD             7223 non-null   int64  
 4   LEN_OF_RLTN_IN_MNTH        7223 non-null   int64  
 5   NO_OF_L_CR_TXNS            7223 non-null   int64  
 6   NO_OF_BR_CSH_WDL_DR_TXNS   7223 non-null   int64  
 7   NO_OF_ATM_DR_TXNS          7223 non-null   int64  
 8   NO_OF_NET_DR_TXNS          7223 non-null   int64  
 9   NO_OF_MOB_DR_TXNS          7223 non-null   int64  
 10  AMT_ATM_DR                 7223 non-null   int64  
 11  AMT_BR_CSH_WDL_DR          7223 non-null   int64  
 12  AMT_CHQ_DR                 7223 non-null   int64  
 13  AMT_NET_DR                 7223 non-null   float64
 14  AMT_MOB_DR                 7223 non-null   int64  
 15  BALANCE_log                7223 non-null   float64
 16  NO_OF_CHQ_DR_TXNS_log      7223 non-null   float64
 17  GENDER_M                   7223 non-null   uint8  
 18  GENDER_O                   7223 non-null   uint8  
 19  OCCUPATION_SAL             7223 non-null   uint8  
 20  OCCUPATION_SELF-EMP        7223 non-null   uint8  
 21  OCCUPATION_SENP            7223 non-null   uint8  
 22  ACC_TYPE_SA                7223 non-null   uint8  
 23  FLG_HAS_CC_1               7223 non-null   uint8  
 24  FLG_HAS_ANY_CHGS_1         7223 non-null   uint8  
 25  FLG_HAS_NOMINEE_1          7223 non-null   uint8  
 26  FLG_HAS_OLD_LOAN_1         7223 non-null   uint8  
 27  ACC_OP_DATE_MONTH          7223 non-null   int64  
 28  ACC_OP_DATE_DAYNAME        7223 non-null   object 
 29  ACC_OP_DATE_AGE_IN_MONTHS  7223 non-null   int64  
dtypes: float64(3), int64(16), object(1), uint8(10)
memory usage: 1.2+ MB


day_map = {
    'Monday': 1,
    'Tuesday': 2,
    'Wednesday': 3, 
    'Thursday': 4, 
    'Friday': 5,
    'Saturday': 6, 
    'Sunday': 7, 
}
main_data['ACC_OP_DATE_DAYNUM'] = main_data['ACC_OP_DATE_DAYNAME'].replace(day_map)
main_data.drop(['ACC_OP_DATE_DAYNAME'], axis=1, inplace=True)
main_data['ACC_OP_DATE_DAYNUM']
main_data.info()

<class 'pandas.core.frame.DataFrame'>
Int64Index: 7223 entries, 0 to 19999
Data columns (total 30 columns):
 #   Column                     Non-Null Count  Dtype  
---  ------                     --------------  -----  
 0   TARGET                     7223 non-null   int64  
 1   AGE                        7223 non-null   int64  
 2   SCR                        7223 non-null   int64  
 3   HOLDING_PERIOD             7223 non-null   int64  
 4   LEN_OF_RLTN_IN_MNTH        7223 non-null   int64  
 5   NO_OF_L_CR_TXNS            7223 non-null   int64  
 6   NO_OF_BR_CSH_WDL_DR_TXNS   7223 non-null   int64  
 7   NO_OF_ATM_DR_TXNS          7223 non-null   int64  
 8   NO_OF_NET_DR_TXNS          7223 non-null   int64  
 9   NO_OF_MOB_DR_TXNS          7223 non-null   int64  
 10  AMT_ATM_DR                 7223 non-null   int64  
 11  AMT_BR_CSH_WDL_DR          7223 non-null   int64  
 12  AMT_CHQ_DR                 7223 non-null   int64  
 13  AMT_NET_DR                 7223 non-null   float64
 14  AMT_MOB_DR                 7223 non-null   int64  
 15  BALANCE_log                7223 non-null   float64
 16  NO_OF_CHQ_DR_TXNS_log      7223 non-null   float64
 17  GENDER_M                   7223 non-null   uint8  
 18  GENDER_O                   7223 non-null   uint8  
 19  OCCUPATION_SAL             7223 non-null   uint8  
 20  OCCUPATION_SELF-EMP        7223 non-null   uint8  
 21  OCCUPATION_SENP            7223 non-null   uint8  
 22  ACC_TYPE_SA                7223 non-null   uint8  
 23  FLG_HAS_CC_1               7223 non-null   uint8  
 24  FLG_HAS_ANY_CHGS_1         7223 non-null   uint8  
 25  FLG_HAS_NOMINEE_1          7223 non-null   uint8  
 26  FLG_HAS_OLD_LOAN_1         7223 non-null   uint8  
 27  ACC_OP_DATE_MONTH          7223 non-null   int64  
 28  ACC_OP_DATE_AGE_IN_MONTHS  7223 non-null   int64  
 29  ACC_OP_DATE_DAYNUM         7223 non-null   int64  
dtypes: float64(3), int64(17), uint8(10)
memory usage: 1.2 MB


categorial_data_column_names = ['ACC_OP_DATE_MONTH', 'ACC_OP_DATE_DAYNUM', 'ACC_OP_DATE_AGE_IN_MONTHS']       # A list of categorial columns.
main_data[categorial_data_column_names].head(10)


fig, axs = plt.subplots(
    nrows=1,                                                                # Number of rows of the grid
    ncols=3,                                                                # Number of columns of the grid.
    figsize=(15,5),                                                         # The size each figure
    constrained_layout=True)                                                # To avoid overlap between values

'''
Input:
axis, column/feature name

Output:
display a counplot for number of product purchses for specific values on that feature.

Retunrs:
N/A
''' 
def plot_cartegorial_charts(ax, df_column_name):                            
    ax = sns.countplot(                                                     
        data=main_data[main_data['TARGET'] == 1], 
        x=main_data[df_column_name],
        ax=ax)                                                              
    ax.set_title(df_column_name + ' Profile', fontsize=14)                  

    
column_names_index = 0;

for ax in axs.flat:
    plot_cartegorial_charts(ax, categorial_data_column_names[column_names_index])
    column_names_index = column_names_index + 1


'''
Input:
axis, column/feature name

Output:
display a counplot for number of product purchses for specific values on that feature.

Retunrs:
N/A
''' 
def plot_cartegorial_charts(ax, df_column_name):                                     
    ax = sns.countplot(                                                   
        x=df_column_name,                                                   
        data=main_data[main_data['TARGET'] == 1], # select on rows that led to product purchase                                                   
        ax = ax)                                                        

    ax.set_xlabel(df_column_name)                                                                               
    ax.set_title(df_column_name + ' Profile', fontsize=14)                  


'''
Input:
N/A

Output:
Goes through all the features and displasy a counplot for number of product purchses for specific values on that feature.

Retunrs:
N/A
''' 
def PlotCountPlotGrid():
    col_names = main_data.columns.tolist()
    print(col_names)

    col_index = 0
    for r in range(0, int(len(col_names)), 3):
        fig, box = plt.subplots(
            nrows=1,
            ncols=3, 
            figsize=(15,3),
            constrained_layout=True)

        for index in range(0, 3):
            if col_index < int(len(col_names)):
                plot_cartegorial_charts(box.flat[index], col_names[col_index])
            col_index += 1
            
PlotCountPlotGrid()

['TARGET', 'AGE', 'SCR', 'HOLDING_PERIOD', 'LEN_OF_RLTN_IN_MNTH', 'NO_OF_L_CR_TXNS', 'NO_OF_BR_CSH_WDL_DR_TXNS', 'NO_OF_ATM_DR_TXNS', 'NO_OF_NET_DR_TXNS', 'NO_OF_MOB_DR_TXNS', 'AMT_ATM_DR', 'AMT_BR_CSH_WDL_DR', 'AMT_CHQ_DR', 'AMT_NET_DR', 'AMT_MOB_DR', 'BALANCE_log', 'NO_OF_CHQ_DR_TXNS_log', 'GENDER_M', 'GENDER_O', 'OCCUPATION_SAL', 'OCCUPATION_SELF-EMP', 'OCCUPATION_SENP', 'ACC_TYPE_SA', 'FLG_HAS_CC_1', 'FLG_HAS_ANY_CHGS_1', 'FLG_HAS_NOMINEE_1', 'FLG_HAS_OLD_LOAN_1', 'ACC_OP_DATE_MONTH', 'ACC_OP_DATE_AGE_IN_MONTHS', 'ACC_OP_DATE_DAYNUM']


PlotHistBoxGrid()

['TARGET', 'AGE', 'SCR', 'HOLDING_PERIOD', 'LEN_OF_RLTN_IN_MNTH', 'NO_OF_L_CR_TXNS', 'NO_OF_BR_CSH_WDL_DR_TXNS', 'NO_OF_ATM_DR_TXNS', 'NO_OF_NET_DR_TXNS', 'NO_OF_MOB_DR_TXNS', 'AMT_ATM_DR', 'AMT_BR_CSH_WDL_DR', 'AMT_CHQ_DR', 'AMT_NET_DR', 'AMT_MOB_DR', 'BALANCE_log', 'NO_OF_CHQ_DR_TXNS_log', 'ACC_OP_DATE_MONTH', 'ACC_OP_DATE_AGE_IN_MONTHS', 'ACC_OP_DATE_DAYNUM']


fig, ax = plt.subplots(figsize=(18,18)) 
sns.heatmap(data=main_data.corr(), annot=True, linewidths=.5, ax=ax)
plt.show()


main_data.corr()[abs(main_data.corr()['TARGET']) < 0.05]


main_data.corr()[abs(main_data.corr()['TARGET']) < 0.05].index.to_list()

['AGE',
 'LEN_OF_RLTN_IN_MNTH',
 'GENDER_O',
 'FLG_HAS_NOMINEE_1',
 'FLG_HAS_OLD_LOAN_1',
 'ACC_OP_DATE_MONTH',
 'ACC_OP_DATE_AGE_IN_MONTHS',
 'ACC_OP_DATE_DAYNUM']


# drop features with corrolation of less than 0.05 to the tagert variable
list_of_features_to_drop = main_data.corr()[abs(main_data.corr()['TARGET']) < 0.05].index.to_list()
main_data.drop(list_of_features_to_drop, axis=1, inplace=True)
main_data.info()

<class 'pandas.core.frame.DataFrame'>
Int64Index: 7223 entries, 0 to 19999
Data columns (total 22 columns):
 #   Column                    Non-Null Count  Dtype  
---  ------                    --------------  -----  
 0   TARGET                    7223 non-null   int64  
 1   SCR                       7223 non-null   int64  
 2   HOLDING_PERIOD            7223 non-null   int64  
 3   NO_OF_L_CR_TXNS           7223 non-null   int64  
 4   NO_OF_BR_CSH_WDL_DR_TXNS  7223 non-null   int64  
 5   NO_OF_ATM_DR_TXNS         7223 non-null   int64  
 6   NO_OF_NET_DR_TXNS         7223 non-null   int64  
 7   NO_OF_MOB_DR_TXNS         7223 non-null   int64  
 8   AMT_ATM_DR                7223 non-null   int64  
 9   AMT_BR_CSH_WDL_DR         7223 non-null   int64  
 10  AMT_CHQ_DR                7223 non-null   int64  
 11  AMT_NET_DR                7223 non-null   float64
 12  AMT_MOB_DR                7223 non-null   int64  
 13  BALANCE_log               7223 non-null   float64
 14  NO_OF_CHQ_DR_TXNS_log     7223 non-null   float64
 15  GENDER_M                  7223 non-null   uint8  
 16  OCCUPATION_SAL            7223 non-null   uint8  
 17  OCCUPATION_SELF-EMP       7223 non-null   uint8  
 18  OCCUPATION_SENP           7223 non-null   uint8  
 19  ACC_TYPE_SA               7223 non-null   uint8  
 20  FLG_HAS_CC_1              7223 non-null   uint8  
 21  FLG_HAS_ANY_CHGS_1        7223 non-null   uint8  
dtypes: float64(3), int64(12), uint8(7)
memory usage: 1.2 MB


fig, ax = plt.subplots(figsize=(18,18)) 
sns.heatmap(data=main_data.corr(), annot=True, linewidths=.5, ax=ax)
plt.show()


PlotHistBoxGrid()

['TARGET', 'SCR', 'HOLDING_PERIOD', 'NO_OF_L_CR_TXNS', 'NO_OF_BR_CSH_WDL_DR_TXNS', 'NO_OF_ATM_DR_TXNS', 'NO_OF_NET_DR_TXNS', 'NO_OF_MOB_DR_TXNS', 'AMT_ATM_DR', 'AMT_BR_CSH_WDL_DR', 'AMT_CHQ_DR', 'AMT_NET_DR', 'AMT_MOB_DR', 'BALANCE_log', 'NO_OF_CHQ_DR_TXNS_log']


sns.pairplot(main_data, kind="reg", hue="TARGET")

<seaborn.axisgrid.PairGrid at 0x209c4cdaee0>


X = main_data.drop("TARGET", axis=1)
Y = main_data.pop("TARGET")
x_train, x_test, y_train, y_test = train_test_split(X, Y, test_size=0.5, random_state=1)


feature_names = list(X.columns)
print(feature_names)

['SCR', 'HOLDING_PERIOD', 'NO_OF_L_CR_TXNS', 'NO_OF_BR_CSH_WDL_DR_TXNS', 'NO_OF_ATM_DR_TXNS', 'NO_OF_NET_DR_TXNS', 'NO_OF_MOB_DR_TXNS', 'AMT_ATM_DR', 'AMT_BR_CSH_WDL_DR', 'AMT_CHQ_DR', 'AMT_NET_DR', 'AMT_MOB_DR', 'BALANCE_log', 'NO_OF_CHQ_DR_TXNS_log', 'GENDER_M', 'OCCUPATION_SAL', 'OCCUPATION_SELF-EMP', 'OCCUPATION_SENP', 'ACC_TYPE_SA', 'FLG_HAS_CC_1', 'FLG_HAS_ANY_CHGS_1']


# Functions to calculate different metrics

def DisplayConfusionMatrix(model, y_actual, labels=[1,0]):
    y_predict = model.predict(x_test)
    confusion_matrix = metrics.confusion_matrix(y_actual, y_predict, labels=[0, 1])
    confusion_matix_df = pd.DataFrame(
        confusion_matrix,
        index   = [i for i in ["Actual No", "Actual Yes"]],
        columns = [i for i in ["Predicted - No", "Predicted - Yes"]])
    
    group_counts = ["{0:0.0f}".format(value) for value in confusion_matrix.flatten()]
    group_percetages = ["{0:.2%}".format(value) for value in confusion_matrix.flatten()/np.sum(confusion_matrix)]
    
    labels = [f"{v1}\n{v2}" for v1, v2 in zip(group_counts, group_percetages)]
    labels = np.array(labels).reshape(2, 2)
    plt.figure(figsize= (10,7))
    sns.heatmap(confusion_matix_df, annot=labels,fmt='')
    plt.ylabel("True Label")
    plt.xlabel("Predicted Label")
    
def GetMetricsScore(model):
    
    pred_train = model.predict(x_train)
    pred_test  = model.predict(x_test)
    
    train_accuracy = model.score(x_train, y_train)
    test_accuracy  = model.score(x_test, y_test)
    
    train_recall = metrics.recall_score(y_train, pred_train)
    test_recall  = metrics.recall_score(y_test, pred_test)
    
    train_precision = metrics.precision_score(y_train, pred_train)
    test_precision  = metrics.precision_score(y_test, pred_test)

    f1_score_train = 2 * ((train_precision * train_recall)/(train_precision + train_recall))
    f1_score_test = 2 * ((test_precision * test_recall)/(test_precision + test_recall))
    
    return {
        'Accuracy_Test'   : test_accuracy,
        'Accuracy_Train'  : train_accuracy,
        'Recall_Test'     : test_recall,
        'Recall_Train'    : train_recall,
        'Precision_Test'  : test_precision,
        'Precision_Train' : train_precision,
        'F1_Score_Train'  : f1_score_train,
        'F1_Score_Test'   : f1_score_test
    }
        

def DisplayImportance(model):
    gini_importances = pd.DataFrame(model.feature_importances_, columns=["Imp"], index=x_train.columns).sort_values(by = "Imp")
    print(gini_importances)
    
def DisplayImportanceChart(model):
    importances = model.feature_importances_
    indices = np.argsort(importances)

    plt.figure(figsize=(12, 12))
    plt.title('Feature Importance')
    plt.barh(range(len(indices)), importances[indices], color='violet', align='center')
    plt.yticks(range(len(indices)), [list(X.columns)[i] for i in indices])
    plt.xlabel('Relative Importance')
    plt.show()
    
def PlotTree(model, size = [20, 30]):
    plt.figure(figsize=(size[0],size[1]))
    tree.plot_tree(model, feature_names=feature_names, filled=True, fontsize=9, node_ids=True, class_names=True)
    plt.show()
    
def PrintCoeficients(model):
    coefs = pd.DataFrame(
        np.append(model.coef_, model.intercept_),
        index=x_train.columns.tolist() + ["Intercept"],
        columns=["Coefficients"],
    )

    coefs.sort_values('Coefficients')


dTree_Model = DecisionTreeClassifier(criterion= 'gini', random_state=1)
dTree_Model.fit(x_train, y_train)

DecisionTreeClassifier(random_state=1)


dTree_Model.get_params()

{'ccp_alpha': 0.0,
 'class_weight': None,
 'criterion': 'gini',
 'max_depth': None,
 'max_features': None,
 'max_leaf_nodes': None,
 'min_impurity_decrease': 0.0,
 'min_samples_leaf': 1,
 'min_samples_split': 2,
 'min_weight_fraction_leaf': 0.0,
 'random_state': 1,
 'splitter': 'best'}


PlotTree(dTree_Model)


DisplayConfusionMatrix(dTree_Model, y_test)


dTree_model_score = GetMetricsScore(dTree_Model)
pprint.pprint(dTree_model_score)

{'Accuracy_Test': 0.758859357696567,
 'Accuracy_Train': 1.0,
 'F1_Score_Test': 0.6840768951759159,
 'F1_Score_Train': 1.0,
 'Precision_Test': 0.6253315649867374,
 'Precision_Train': 1.0,
 'Recall_Test': 0.755004003202562,
 'Recall_Train': 1.0}


DisplayImportanceChart(dTree_Model)


# choose the type of classifier
dTree_Model = DecisionTreeClassifier(random_state=1, criterion= 'gini')

# Grid of params to choose from
parameters = {
    "max_depth": np.arange(1, 10),
    "min_samples_leaf": [1, 2, 5, 7, 10, 15, 20],
    "max_leaf_nodes" : [2, 3, 5, 10],
    "min_impurity_decrease": [0.001, 0.01, 0.1]
}

# type of score used to compare paramter combinations
acc_scorer =metrics.make_scorer(metrics.recall_score)

# run the grid search
grid_obj = GridSearchCV(dTree_Model, parameters, scoring=acc_scorer, cv=5)
grid_obj = grid_obj.fit(x_train, y_train)

# set the clf to the best combination of parameters
dTree_Model = grid_obj.best_estimator_

# fit the best algorithm to the data
dTree_Model.fit(x_train, y_train)

DecisionTreeClassifier(max_depth=6, max_leaf_nodes=10,
                       min_impurity_decrease=0.001, random_state=1)


PlotTree(dTree_Model, [15, 10])


dTree_model_score = GetMetricsScore(dTree_Model)
pprint.pprint(dTree_model_score)

{'Accuracy_Test': 0.6802325581395349,
 'Accuracy_Train': 0.7009138742730545,
 'F1_Score_Test': 0.3492957746478873,
 'F1_Score_Train': 0.4026548672566372,
 'Precision_Test': 0.5893536121673004,
 'Precision_Train': 0.6678899082568808,
 'Recall_Test': 0.24819855884707767,
 'Recall_Train': 0.28820269200316706}


DisplayConfusionMatrix(dTree_Model, y_test)


DisplayImportanceChart(dTree_Model)


# fit model on train
logistic_reg_model = LogisticRegression()
logistic_reg_model.fit(x_train,y_train)

# predict on test
y_predict = logistic_reg_model.predict(x_test)


logistic_reg_model.get_params()

{'C': 1.0,
 'class_weight': None,
 'dual': False,
 'fit_intercept': True,
 'intercept_scaling': 1,
 'l1_ratio': None,
 'max_iter': 100,
 'multi_class': 'auto',
 'n_jobs': None,
 'penalty': 'l2',
 'random_state': None,
 'solver': 'lbfgs',
 'tol': 0.0001,
 'verbose': 0,
 'warm_start': False}


logistic_reg_model_score = GetMetricsScore(logistic_reg_model)
pprint.pprint(logistic_reg_model_score)

{'Accuracy_Test': 0.647563676633444,
 'Accuracy_Train': 0.6477430074771532,
 'F1_Score_Test': 0.16524590163934424,
 'F1_Score_Train': 0.15984147952443856,
 'Precision_Test': 0.45652173913043476,
 'Precision_Train': 0.4820717131474104,
 'Recall_Test': 0.10088070456365092,
 'Recall_Train': 0.0958036421219319}


DisplayConfusionMatrix(logistic_reg_model, y_test)

	count	mean	std	min	25%	50%	75%	max
TARGET	20000.0	0.125600	0.331406	0.0	0.000	0.000	0.00	1.00
AGE	20000.0	38.418150	9.577985	21.0	30.000	38.000	46.00	55.00
BALANCE	20000.0	511362.192118	723592.961792	0.0	64754.025	231675.845	653876.85	8360430.57
SCR	20000.0	440.150300	249.887582	100.0	227.000	364.000	644.00	999.00
HOLDING_PERIOD	20000.0	14.955650	8.642634	1.0	7.000	15.000	22.00	31.00
LEN_OF_RLTN_IN_MNTH	20000.0	125.239300	54.953936	29.0	79.000	125.000	172.00	221.00
NO_OF_L_CR_TXNS	20000.0	12.348050	12.151921	0.0	6.000	10.000	14.00	75.00
NO_OF_BR_CSH_WDL_DR_TXNS	20000.0	1.883000	2.263596	0.0	1.000	1.000	2.00	15.00
NO_OF_ATM_DR_TXNS	20000.0	1.028950	1.497239	0.0	0.000	1.000	1.00	25.00
NO_OF_NET_DR_TXNS	20000.0	1.172450	2.404246	0.0	0.000	0.000	1.00	22.00
NO_OF_MOB_DR_TXNS	20000.0	0.411750	2.009854	0.0	0.000	0.000	0.00	25.00
NO_OF_CHQ_DR_TXNS	20000.0	2.137550	2.384893	0.0	0.000	2.000	4.00	15.00
AMT_ATM_DR	20000.0	10990.015000	15304.891499	0.0	0.000	6900.000	15800.00	199300.00
AMT_BR_CSH_WDL_DR	20000.0	378474.478000	331417.730129	0.0	2990.000	340150.000	674675.00	999930.00
AMT_CHQ_DR	20000.0	124520.005150	439379.541425	0.0	0.000	23840.000	72470.00	4928640.00
AMT_NET_DR	20000.0	237307.771646	319067.664858	0.0	0.000	0.000	473970.50	999854.00
AMT_MOB_DR	20000.0	22424.704400	49484.803631	0.0	0.000	0.000	0.00	199667.00

	TARGET	AGE	SCR	HOLDING_PERIOD	LEN_OF_RLTN_IN_MNTH	NO_OF_L_CR_TXNS	NO_OF_BR_CSH_WDL_DR_TXNS	NO_OF_ATM_DR_TXNS	NO_OF_NET_DR_TXNS	NO_OF_MOB_DR_TXNS	...	OCCUPATION_SELF-EMP	OCCUPATION_SENP	ACC_TYPE_SA	FLG_HAS_CC_1	FLG_HAS_ANY_CHGS_1	FLG_HAS_NOMINEE_1	FLG_HAS_OLD_LOAN_1	ACC_OP_DATE_MONTH	ACC_OP_DATE_AGE_IN_MONTHS	ACC_OP_DATE_DAYNUM
AGE	-0.009223	1.000000	-0.029402	0.013523	-0.018698	0.005099	0.015904	-0.027290	-0.016679	-0.014019	...	-0.027127	0.006164	-0.016268	-0.014591	-0.025913	0.015852	0.005460	0.003891	-0.020545	0.002734
LEN_OF_RLTN_IN_MNTH	-0.041794	-0.018698	-0.006686	0.001693	1.000000	-0.001055	-0.021999	-0.021472	-0.034341	-0.007068	...	-0.013797	-0.000761	0.012905	-0.011855	0.035218	0.009722	0.010737	-0.004907	0.473340	0.002585
GENDER_O	0.038179	-0.004135	0.018311	0.009005	0.026286	0.156378	0.046234	0.032514	0.071129	0.035202	...	-0.024988	-0.060924	-0.195260	-0.009468	0.001352	-0.022695	0.035103	-0.017569	-0.006944	0.030654
FLG_HAS_NOMINEE_1	0.019112	0.015852	-0.007890	-0.000047	0.009722	0.006320	0.015547	-0.007666	-0.000926	-0.018605	...	0.006726	-0.006132	-0.027974	0.008756	-0.011675	1.000000	0.009486	0.000225	0.030413	-0.020522
FLG_HAS_OLD_LOAN_1	-0.017296	0.005460	0.013985	0.001182	0.010737	0.010725	-0.005594	-0.013546	0.003095	-0.015268	...	-0.019480	-0.016632	-0.022605	-0.021923	-0.015369	0.009486	1.000000	-0.021037	-0.003534	0.029263
ACC_OP_DATE_MONTH	-0.002745	0.003891	-0.018120	0.027016	-0.004907	0.012051	-0.002720	0.006753	-0.011249	-0.001342	...	0.005159	-0.000500	-0.013417	0.006846	0.030880	0.000225	-0.021037	1.000000	-0.028531	-0.001258
ACC_OP_DATE_AGE_IN_MONTHS	-0.038102	-0.020545	0.005769	0.019832	0.473340	0.005860	-0.017459	-0.019942	-0.011621	-0.014547	...	0.003848	0.008427	0.002293	-0.002313	0.027237	0.030413	-0.003534	-0.028531	1.000000	0.000490
ACC_OP_DATE_DAYNUM	0.003966	0.002734	0.000867	-0.003369	0.002585	0.023907	0.002206	-0.017069	-0.007002	-0.016089	...	-0.006226	-0.014863	-0.009728	0.009409	0.011243	-0.020522	0.029263	-0.001258	0.000490	1.000000

Assignment 3: Bancassurance¶

Contents:¶

1. Data Overview and EDA:

2. Outlier Treatment and Feature Engineering:

3. Building Decision Tree Model and Analysis:

4. Building Logistic Regression Model and Analysis:

5. Conclusions

	CUST_ID	AGE	GENDER	BALANCE	OCCUPATION	SCR	HOLDING_PERIOD	ACC_TYPE	ACC_OP_DATE	...	NO_OF_CHQ_DR_TXNS	FLG_HAS_CC	AMT_ATM_DR	AMT_BR_CSH_WDL_DR	AMT_CHQ_DR	AMT_NET_DR	AMT_MOB_DR	FLG_HAS_ANY_CHGS	FLG_HAS_NOMINEE	FLG_HAS_OLD_LOAN
0	C7927	27	M	3383.75	SELF-EMP	776	30	SA	3/23/2005	...	0	0	13100	0	0	973557.0	0	0	1	1
1	C6877	47	M	287489.04	SAL	324	28	SA	10-11-2008	...	0	0	6600	0	0	799813.0	0	1	1	0
2	C19922	40	M	18216.88	SELF-EMP	603	2	SA	4/26/2012	...	2	0	11200	561120	49320	997570.0	0	1	1	1
3	C8183	53	M	71720.48	SAL	196	13	CA	07-04-2008	...	4	0	26100	673590	60780	741506.0	71388	0	1	0
4	C12123	36	M	1671622.89	PROF	167	24	SA	12/29/2001	...	0	0	0	808480	0	0.0	0	0	1	0
5	C257	42	F	521685.69	PROF	493	26	SA	06-07-2001	...	0	1	18500	379310	0	0.0	0	0	1	1
6	C8300	30	M	204458.60	PROF	479	14	SA	8/25/2002	...	4	0	6200	0	10580	770065.0	0	1	0	1
7	C5999	53	F	58516.34	SAL	562	25	CA	3/15/2009	...	0	0	0	945160	0	0.0	0	0	1	1
8	C5692	42	F	13158.14	SELF-EMP	105	15	SA	02-08-2010	...	1	1	35400	198430	51490	326421.0	170332	0	1	1
9	C3241	30	M	831150.18	PROF	170	13	SA	03-02-2008	...	4	0	18000	869880	32610	0.0	0	0	0	0

	Non-Nulls	Nulls %	Type
ACC_OP_DATE	20000	0.00%	object
ACC_TYPE	20000	0.00%	object
AGE	20000	0.00%	int64
AMT_ATM_DR	20000	0.00%	int64
AMT_BR_CSH_WDL_DR	20000	0.00%	int64
AMT_CHQ_DR	20000	0.00%	int64
AMT_MOB_DR	20000	0.00%	int64
AMT_NET_DR	20000	0.00%	float64
BALANCE	20000	0.00%	float64
FLG_HAS_ANY_CHGS	20000	0.00%	int64
FLG_HAS_CC	20000	0.00%	int64
FLG_HAS_NOMINEE	20000	0.00%	int64
FLG_HAS_OLD_LOAN	20000	0.00%	int64
GENDER	20000	0.00%	object
HOLDING_PERIOD	20000	0.00%	int64
LEN_OF_RLTN_IN_MNTH	20000	0.00%	int64
NO_OF_ATM_DR_TXNS	20000	0.00%	int64
NO_OF_BR_CSH_WDL_DR_TXNS	20000	0.00%	int64
NO_OF_CHQ_DR_TXNS	20000	0.00%	int64
NO_OF_L_CR_TXNS	20000	0.00%	int64
NO_OF_MOB_DR_TXNS	20000	0.00%	int64
NO_OF_NET_DR_TXNS	20000	0.00%	int64
OCCUPATION	20000	0.00%	object
SCR	20000	0.00%	int64
TARGET	20000	0.00%	int64

	AGE	GENDER	BALANCE	OCCUPATION	SCR	HOLDING_PERIOD	ACC_TYPE	ACC_OP_DATE	LEN_OF_RLTN_IN_MNTH	...	NO_OF_CHQ_DR_TXNS	FLG_HAS_CC	AMT_ATM_DR	AMT_BR_CSH_WDL_DR	AMT_CHQ_DR	AMT_NET_DR	AMT_MOB_DR	FLG_HAS_ANY_CHGS	FLG_HAS_NOMINEE	FLG_HAS_OLD_LOAN
0	27	M	3383.75	SELF-EMP	776	30	SA	2005-03-23	146	...	0	0	13100	0	0	973557.0	0	0	1	1
1	47	M	287489.04	SAL	324	28	SA	2008-10-11	104	...	0	0	6600	0	0	799813.0	0	1	1	0
2	40	M	18216.88	SELF-EMP	603	2	SA	2012-04-26	61	...	2	0	11200	561120	49320	997570.0	0	1	1	1
3	53	M	71720.48	SAL	196	13	CA	2008-07-04	107	...	4	0	26100	673590	60780	741506.0	71388	0	1	0
4	36	M	1671622.89	PROF	167	24	SA	2001-12-29	185	...	0	0	0	808480	0	0.0	0	0	1	0
6	30	M	204458.60	PROF	479	14	SA	2002-08-25	177	...	4	0	6200	0	10580	770065.0	0	1	0	1
7	53	F	58516.34	SAL	562	25	CA	2009-03-15	99	...	0	0	0	945160	0	0.0	0	0	1	1
8	42	F	13158.14	SELF-EMP	105	15	SA	2010-02-08	88	...	1	1	35400	198430	51490	326421.0	170332	0	1	1
10	52	M	13683.16	SELF-EMP	395	7	SA	2004-01-26	160	...	2	0	11300	632710	13080	381666.0	0	0	1	1
11	43	M	272607.05	PROF	164	24	CA	2002-03-06	183	...	0	1	8900	362780	0	0.0	171334	0	1	0

	ACC_OP_DATE_MONTH	ACC_OP_DATE_DAYNUM	ACC_OP_DATE_AGE_IN_MONTHS
0	3	3	203
1	10	6	160
2	4	4	118
3	7	5	163
4	12	6	242
6	8	7	234
7	3	7	155
8	2	1	144
10	1	1	217
11	3	3	239

	ACC_OP_DATE_MONTH	ACC_OP_DATE_DAYNUM	ACC_OP_DATE_AGE_IN_MONTHS
0	3	3	203
1	10	6	160
2	4	4	118
3	7	5	163
4	12	6	242
6	8	7	234
7	3	7	155
8	2	1	144
10	1	1	217
11	3	3	239

	ACC_OP_DATE_MONTH	ACC_OP_DATE_DAYNUM	ACC_OP_DATE_AGE_IN_MONTHS
0	3	3	203
1	10	6	160
2	4	4	118
3	7	5	163
4	12	6	242
6	8	7	234
7	3	7	155
8	2	1	144
10	1	1	217
11	3	3	239